Gender diversity in corporate boardroom and tax avoidance the evidence in hose listed firms

UNIVERSITY OF ECONOMICS ERASMUS UNVERSITY ROTTERDAM INSTITUTE OF SOCIAL STUDIES HO CHI MINH CITY THE NETHERLANDS VIETNAM VIETNAM – THE NETHERLANDS PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS GENDER DIVERSITY IN CORPORATE BOARDROOM AND TAX AVOIDANCE THE EVIDENCE IN HOSE LISTED FIRMS BY DAO THI HAN MASTER OF ARTS IN DEVELOPMENT ECONOMICS HO CHI MINH CITY, December 2016 UNIVERSITY OF ECONOMICS i ERASMUS UNVERSITY ROTTERDAM HO CHI MINH CITY INSTITUTE OF SOCIAL STUDIES VIETNAM THE NETHERLANDS VIETNAM – THE NETHERLANDS PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS GENDER DIVERSITY IN CORPORATE BOARDROOM AND TAX AVOIDANCE THE EVIDENCE IN HOSE LISTED FIRMS A thesis submitted in partial fulfilment of the requirements for the degree of MASTER OF ARTS IN DEVELOPMENT ECONOMICS BY DAO THI HAN Academic Supervisor: Nguyen Thi Thuy Linh HO CHI MINH CITY, December 2016 ii ABSTRACT Using data set of 296 publicly listed firms in Ho Chi Minh Stock Exchange from 2010 to 2015, the study analyses the engagement of female board members, also in case of being leader of board and executive manager, on tax avoidance activities, measured by three proxies The fixed effect regression results indicate gender diversity in boardroom is negatively associated with tax avoidance measured by effective tax rate but chairwomen are more engaged in tax avoidance measured by book-tax difference As a result, the presence of women in boardroom of HOSE listed firms is important to shareholders who consider about firms’ transparency or profit iii TABLE OF CONTENT LIST OF ABBREVIATIONS iv LIST OF FIGURE v LIST OF TABLE vi CHAPTER ONE INTRODUCTION .1 1.1 Vietnam overview .1 1.1.1 Female labor and pay gap in Vietnam 1.1.2 Women participation in firm management in Vietnam 1.1.3 Tax avoidance in Vietnam 1.2 Research objective 1.3 Research design CHAPTER TWO LITERATURE REVIEW 2.1 Gender diversity and corporate governance .9 2.1.1 Resource-dependence theory 11 2.1.2 Agency theory 11 2.1.3 Gender equality reaction in corporate boardroom 12 2.2 Tax avoidance and corporate governance 13 2.2.1 Tax avoidance 13 2.2.2 Tax avoidance and corporate governance 14 2.3 Gender diversity in boardroom and tax avoidance 15 2.3.1 Women’ participation in boardroom and tax avoidance 16 2.3.2 Chairwomen and tax avoidance .17 2.3.3 Female executive in boardroom and tax avoidance .18 i 2.3.4 Summary 19 CHAPTER THREE METHODOLOGY .20 3.1 Analytical framework .20 3.2 Data and data source 20 3.3 Research model 21 3.3.1 Baseline model 21 3.3.2 Variable explanation 23 3.4 Research methodology 26 3.4.1 Regression models 26 3.4.2 Robust standard Errors 27 CHAPTER FOUR EMPIRICAL RESULT 28 4.1 Descriptive statistic 28 4.1.1 Summary descriptive statistic 28 4.1.2 Women board directors in HOSE listed firms .30 4.1.3 Tax expense in HOSE listed firms having women board directors .32 4.2 Empirical result 33 4.2.1 GAAP effective tax rate 34 4.2.2 CASH effective tax rate 35 4.2.3 Book-tax differences 37 4.3 Summary results 38 CHAPTER FIVE CONCLUSION .40 4.1 Conclusions 40 4.2 Implications 41 4.3 Limitations 41 ii REFERRENCE 43 APPENDIX A SELECTED FIRMS LISTED IN HOSE 48 APPENDIX B CORRELATION MATRIX 50 APPENDIX C MULTICOLLINEARITY TEST 52 APPENDIX D PANEL DATA REGRESSION RESULTS – FEM .54 APPENDIX E PANEL DATA REGRESSION RESULTS – REM .59 APPENDIX F HETEROSKEDASTICITY AND AUTOCORRELATION TEST .64 APPENDIX G REGRESSION WITH ADJUSTED STANDARD ERRORS 66 iii LIST OF ABBREVIATIONS BTD: book-tax different CEO(s): Chief of Executive officer(s) CFO(s): Chief of Finance officer(s) CIT: Corporate Income Tax ETR: effective tax rate EU: European Union FDI: Foreign directed investment FEM: Fixed effect model GAAP: Generally Accepted Accounting Principle GSO: General Statistics Office HNX: Hanoi Stock Exchange HOSE: Ho Chi Minh Stock Exchange ILO: International Labor Organization OECD: Organization for Economic Co-operation and Development REM: Random effect model ROA: Return on asset ROE: Return on equity ROI: Return on investment SG&A: Selling, general and administration VAT: Value-added tax VCCI: Vietnam Chamber of Commerce and Industry VWEC: Vietnam Women Entrepreneurs Council iv LIST OF FIGURE FIGURE 1.1 LABOR FORCE PARTICIPATION RATES (%) FIGURE 1.2: NATIONWIDE FEMALE EMPLOYED POPULATION .2 FIGURE 2.1: THE SCOPE OF CORPORATE GOVERNANCE .10 FIGURE 3.1: ANALYTICAL FRAMEWORK 20 FIGURE 4.1: GRAPH HISTOGRAM OF TAX AVOIDANCE MEASURES 28 FIGURE 4.2: GRAPH HISTOGRAM OF CORPORATE BOARD SIZE AND WOMEN MEMBERS 30 FIGURE 4.3: HOSE LISTED FIRMS HAVING WOMEN PARTICIPATING BOARDROOM .31 FIGURE 4.4: NUMBER OF EXECUTIVE AND NON-EXECUTIVE FEMALE BOARD DIRECTOR 32 FIGURE 4.5: CURRENT TAX EXPENSE IN HOSE LISTED FIRMS (BILLION VND) 32 FIGURE 4.6: CASH TAX PAID IN FIRMS IN HOSE LISTED FIRMS (BILLION VND) 33 v LIST OF TABLE TABLE 1.1: CORPORATE INCOME TAX SINCE 2013 .4 TABLE 2.1: SUMMARY HYPOTHESES AND RESEARCH QUESTIONS 19 TABLE 3.1: VARIABLE CONSTRUCTION 21 TABLE 3.2: EXPLANATORY VARIABLES 22 TABLE 4.1: SUMMARY STATISTICS OF VARIABLES 29 TABLE 4.2: RESULTS OF FEM WITH TAX AVOIDANCE MEASURED BY GAAPETR 34 TABLE 4.3: RESULTS OF FEM WITH TAX AVOIDANCE MEASURED BY CASHETR 36 TABLE 4.4: RESULTS OF FEM WITH TAX AVOIDANCE MEASURED BY BTD 37 TABLE 4.5: SUMMARY HYPOTHESIS TEST RESULTS 39 vi CHAPTER ONE INTRODUCTION 1.1 Vietnam overview 1.1.1 Female labor and pay gap in Vietnam According to Worldbank data, Vietnam keeps a very high female labor participation rate as high as male’s, not lower than 72 percent since 2000, shown in Figure 1.1 (while male rate is around 82 percent) General Statistics Office (GSO) of Vietnam also reports more than 40 percent of the nationwide labor force are women (see Figure 1.2) However, the International Labor Organization (ILO) identifies that gender pay gap in Vietnam has been widened Vietnamese women earn less than men thirteen percent in 2011 and twenty to thirty percent in 2012 while global average gender pay gap is around 17 percent The latest Labor Force Survey Report (2012) shows that women earn less than male counterparts in all economic sectors, even the favor-female industries like healthcare, social works Hence, the remuneration seems to reflect gender of worker instead of content of work It is clear that the principle of “equal pay for work of equal value” stipulated in the Labor Code need to be implemented Figure 1.1 Labor force participation rates (%) male rate female rate 100 90 80 70 60 50 40 30 20 10 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 (Source: worldbank.org) (female/male) labor force participation rate = % of (female/male) population ages 15+ having a job -F test that all u_i=0: F(289, 1231) = 2.99 Prob > F = 0.0000 Figure D.3: FEM regression result of model Fixed-effects (within) regression Group variable: id R-sq: within = 0.0202 between = 0.0244 overall = 0.0142 corr(u_i, Xb) Number of obs Number of groups Obs per group: = avg = max = F(11,1231) Prob > F = = -0.2566 0.0084 -F test that all u_i=0: F(289, 1231) = 2.98 Prob > F = 0.0000 Figure D.4: FEM regression result of model Fixed-effects (within) regression Group variable: id R-sq: within between overall = corr(u_i, Xb) 55 -F test that all u_i=0: F(289, 1223) = 3.29 Prob > F = 0.0000 Figure D.5: FEM regression result of model Fixed-effects (within) regression Group variable: id R-sq: within = 0.0412 between = 0.0150 overall = 0.0142 corr(u_i, Xb) Number of obs Number of groups Obs per group: = avg = max = F(11,1223) Prob > F = = -0.3637 0.0000 F test that all u_i=0: Figure D.6: FEM regression result of model Fixed-effects (within) regression Group variable: id R-sq: within between = 0.0152 overall = 0.0149 corr(u_i, Xb) 56 -F test that all u_i=0: F(289, 1223) = 3.27 Prob > F = 0.0000 Figure D.7: FEM regression result of model Fixed-effects (within) regression Group variable: id R-sq: within = 0.4209 between = 0.3361 overall = 0.3388 corr(u_i, Xb) Number of obs Number of groups Obs per group: = avg = max = F(11,1322) Prob > F = = -0.3080 0.0000 F test that all u_i=0: Figure D.8: FEM regression result of model Fixed-effects (within) regression Group variable: id R-sq: within between = 0.3370 overall = 0.3395 corr(u_i, Xb) 57 -F test that all u_i=0: F(289, 1322) = 5.42 Prob > F = 0.0000 Figure D.9: FEM regression result of model Fixed-effects (within) regression Group variable: id R-sq: within = 0.4210 between = 0.3365 overall = 0.3390 corr(u_i, Xb) Number of obs Number of groups Obs per group: = avg = max = F(11,1322) Prob > F = = -0.3074 0.0000 -F test that all u_i=0: F(289, 1322) = 5.42 Prob > F = 0.0000 58 APPENDIX E PANEL DATA REGRESSION RESULTS – REM Figure E.1: REM regression result of model Random-effects Group variable: id R-sq: within between corr(u_i, X) Figure E.2: REM regression result of model Random-effects Group variable: id R-sq: within between corr(u_i, X) 59 Figure E.3: REM regression result of model Random-effects Group variable: id R-sq: within between corr(u_i, X) Figure E.4: REM regression result of model Random-effects Group variable: id R-sq: within between corr(u_i, X) 60 Figure E.5: REM regression result of model Random-effects Group variable: id R-sq: within between corr(u_i, X) Figure E.6: REM regression result of model Random-effects Group variable: id R-sq: within between corr(u_i, X) -cashetr | Coef Std Err z P>|z| [95% Conf Interval] 61 Figure E.7: REM regression result of model Random-effects Group variable: id R-sq: within between corr(u_i, X) Figure E.8: REM regression result of model Random-effects Group variable: id R-sq: within between overall 62 corr(u_i, X) = (assumed) Figure E.9: REM regression result of model Random-effects Group variable: id R-sq: within between corr(u_i, X) 63 APPENDIX F HETEROSKEDASTICITY AND AUTOCORRELATION TEST Table F.1: Heteroskedasticity and Autocorrelation test result for model gaapetr wob chairwm wmexe roa size mb slev llev fasset sga1 cash Wooldridge test for autocorrelation in panel data H0: no first-order autocorrelation F( Modified model H0: sigma(i)^2 = sigma^2 for all i chi2 (290) = Prob>chi2 = Table F.2: Heteroskedasticity and Autocorrelation test result for model gaapetr nowob chairwm wmexe roa size mb slev llev fasset sga1 cash Wooldridge test for autocorrelation in panel data H0: no first-order autocorrelation F( Modified model H0: sigma(i)^2 = sigma^2 for all i chi2 (290) = Prob>chi2 = Table F.3: Heteroskedasticity and Autocorrelation test result for model gaapetr womanratio chairwm wmexe roa size mb slev llev fasset sga1 cash Wooldridge test for autocorrelation in panel data H0: no first-order autocorrelation F( 1, Modified Wald test for groupwise heteroskedasticity in fixed effect regression model H0: sigma(i)^2 = sigma^2 for all i chi2 (290) = Prob>chi2 = Table F.4: Heteroskedasticity and Autocorrelation test result for model cashetr wob chairwm wmexe roa size mb slev llev fasset sga1 cash Wooldridge test for autocorrelation in panel data H0: no first-order autocorrelation F( Modified model H0: sigma(i)^2 = sigma^2 for all i chi2 (290) = Prob>chi2 = Table F.5: Heteroskedasticity and Autocorrelation test result for model cashetr nowob chairwm wmexe roa size mb slev llev fasset sga1 cash Wooldridge test for autocorrelation in panel data H0: no first-order autocorrelation F( 1, 256) = 0.036 64 Prob > F = Modified Wald model H0: sigma(i)^2 = sigma^2 for all i chi2 (290) Table F.6: Heteroskedasticity and Autocorrelation test result for model cashetr womanratio chairwm wmexe roa size mb slev llev fasset sga1 cash Wooldridge test for autocorrelation in panel data H0: no first-order autocorrelation F( 1, Prob > F = Wald Modified model H0: sigma(i)^2 = sigma^2 for all i chi2 (290) Prob>chi2 Table F.7: Heteroskedasticity and Autocorrelation test result for model btd wob chairwm wmexe roa size mb slev llev fasset sga1 cash Wooldridge test for autocorrelation in panel data H0: no first-order autocorrelation F( 1, Prob > F = Wald Modified model H0: sigma(i)^2 = sigma^2 for all i chi2 (290) Prob>chi2 Table F.8: Heteroskedasticity and Autocorrelation test result for model btd nowob chairwm wmexe roa size mb slev llev fasset sga1 cash Wooldridge test for autocorrelation in panel data H0: no first-order autocorrelation F( Prob > F = Modified model H0: sigma(i)^2 = sigma^2 for all i chi2 (290) Wald Prob>chi2 = Table F.9: Heteroskedasticity and Autocorrelation test result for model btd womanratio chairwm wmexe roa size mb slev llev fasset sga1 cash Wooldridge test for autocorrelation in panel data H0: no firstorder autocorrelation F( Modified model H0: sigma(i)^2 = sigma^2 for all i chi2 (290) = Prob>chi2 = 65 APPENDIX G REGRESSION WITH ADJUSTED STANDARD ERRORS Figure G.1: Regression Driscoll – Kraay Standard Errors result for model Regression with Driscoll-Kraay standard errors Method: Fixed-effects regression Group variable (i): id maximum lag: Figure G.2: Regression Driscoll – Kraay Standard Errors result for model Regression with Driscoll-Kraay standard errors Method: Fixed-effects regression Group variable (i): id maximum lag: Figure G.3: Regression Driscoll – Kraay Standard Errors result for model Regression with Driscoll-Kraay standard errors Method: Fixed-effects regression Group variable (i): id maximum lag: Number Number F( 11, Prob > within 66 of obs of groups 289) F R-squared = = = = 0.0000 = 0.0202 1532 290 9.05 Figure G.4: Regression Driscoll – Kraay Standard Errors result for model Regression with Driscoll-Kraay standard errors Method: Fixed-effects regression Group variable (i): id maximum lag: Figure G.5: Regression Driscoll – Kraay Standard Errors result for model Regression with Driscoll-Kraay standard errors Method: Fixed-effects regression Group variable (i): id maximum lag: 67 llev fasset sga1 cash _cons Figure G.6: Regression Driscoll – Kraay Standard Errors result for model Regression with Driscoll-Kraay standard errors Method: Fixed-effects regression Group variable (i): id maximum lag: Figure G.7: Regression Driscoll – Kraay Standard Errors result for model Regression with Driscoll-Kraay standard errors Method: Fixed-effects regression Group variable (i): id maximum lag: Figure G.8: Regression Driscoll – Kraay Standard Errors result for model Regression with Driscoll-Kraay standard errors Method: Fixed-effects regression Group variable (i): id Number of obs Number of groups F( 11, 289) 68 = = 1623 = 290 1390.38 | | | | | maximum lag: Figure G.9: Regression Driscoll – Kraay Standard Errors result for model Regression with Driscoll-Kraay standard errors Method: Fixed-effects regression Group variable (i): id maximum lag: 69 ... MINH CITY INSTITUTE OF SOCIAL STUDIES VIETNAM THE NETHERLANDS VIETNAM – THE NETHERLANDS PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS GENDER DIVERSITY IN CORPORATE BOARDROOM AND TAX AVOIDANCE THE. .. 2.2 Tax avoidance and corporate governance 13 2.2.1 Tax avoidance 13 2.2.2 Tax avoidance and corporate governance 14 2.3 Gender diversity in boardroom and tax avoidance. .. and tax avoidance Assuming, all firms may involve in minimizing corporate tax expense if there are few costs associated with tax avoidance However, there are different tax positions between firms